Elements: Software: Towards Efficient Embedded Data Processing

要素：软件：实现高效的嵌入式数据处理

• 批准号: 1835446
• 负责人: Jignesh Patel
• 金额: $ 59.98万
• 依托单位: University of Wisconsin-Madison
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-01-01 至 2023-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1835446&HistoricalAwards=false
• 关键词: Elements; Software; Towards; Efficient; Embedded

Embedded databases are ubiquitous, though that fact may not be widely-recognized. Every smart phone has many embedded databases, which implies that billions of people worldwide carry dozens of databases in their phones/pockets every day. Many of these databases are powered by data processing technology that has not kept up with the pace with which the underling hardware in phones have evolved. As a result, data processing is slow, and consumes more energy than needed. The focus of this proposal is on developing new data processing technology for mobile devices that targets a 10X efficiency and performance improvements. The aims of the project go beyond more efficient data processing on phones to also include more efficient processing in embedded environments, which also includes databases running on laptops. Thus, the project aims for a broad impact on database across a spectrum of mobile devices.The technical contributions of this project are in recognizing that modern hardware, even at the ?low-end? which includes mobile phones and laptops, now have multiple processing cores, relatively large amounts of memory, and flash storage. There is a critical need for a new class of embedded data processing systems that can work efficiently, and effectively on such modern mobile platforms. This project aims to build a system, called Hustle, to address this need. The project will design, develop and implement a range of data processing methods, which include predicate-based concurrency control mechanisms, query processing methods that inherently expose and exploit opportunities for intra and inter-operator parallelism, and query optimization methods that target embedded settings.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

嵌入式数据库无处不在，尽管这一事实可能不会被广泛认可。每个智能手机都有许多嵌入式数据库，这意味着全世界数十亿人每天都在手机/口袋中携带数十个数据库。这些数据库中的许多数据库都由数据处理技术提供动力，这些技术并没有跟上手机中的底座硬件的发展。结果，数据处理缓慢，并且消耗超出所需的能量。该提案的重点是针对以10倍效率和性能提高的移动设备开发新的数据处理技术。该项目的目的超出了手机上更有效的数据处理，还包括在嵌入式环境中更有效的处理，这还包括在笔记本电脑上运行的数据库。因此，该项目旨在对数据库产生广泛的影响。该项目的技术贡献在于认识到现代硬件，即使在“低端”？其中包括手机和笔记本电脑，现在具有多个处理内核，相对较大的内存和闪存存储。至关重要的是，新的嵌入式数据处理系统可以有效地在此类现代移动平台上工作。该项目旨在建立一个称为Hustle的系统，以满足这一需求。该项目将设计，开发和实施一系列数据处理方法，其中包括基于谓词的并发控制机制，查询处理方法，固有地揭示和利用了经营者内部和经营者间并行性的机会，以及针对嵌入的嵌入式设置的查询优化方法，这些奖项反映了NSF的范围，这反映了NSF的范围及其范围的范围。 标准。

项目成果

Simple Adaptive Query Processing vs. Learned Query Optimizers: Observations and Analysis

简单自适应查询处理与学习查询优化器：观察和分析

• DOI: 10.14778/3611479.3611501
• 发表时间: 2023
• 期刊: Proceedings of the VLDB Endowment
• 影响因子: 2.5
• 作者: Zhang, Yunjia;Chronis, Yannis;Patel, Jignesh M.;Rekatsinas, Theodoros
• 通讯作者: Rekatsinas, Theodoros

VIP hashing: adapting to skew in popularity of data on the fly

VIP 哈希：适应动态数据流行度的偏差

• DOI: 10.14778/3547305.3547306
• 发表时间: 2022
• 期刊: Proceedings of the VLDB Endowment
• 影响因子: 2.5
• 作者: Kakaraparthy, Aarati;Patel, Jignesh M.;Kroth, Brian P.;Park, Kwanghyun
• 通讯作者: Park, Kwanghyun

On inter-operator data transfers in query processing

• DOI: 10.1109/icde53745.2022.00066
• 发表时间: 2022-05
• 期刊: 2022 IEEE 38th International Conference on Data Engineering (ICDE)
• 作者: Harshad Deshmukh;Bruhathi Sundarmurthy;Jignesh M. Patel

Database Isolation By Scheduling

• DOI: 10.14778/3461535.3461537
• 发表时间: 2021-05
• 期刊: Proc. VLDB Endow.
• 作者: Kevin P. Gaffney;R. Claus;J. Patel

How Good is My HTAP System?

我的 HTAP 系统有多好？

• DOI: 10.1145/3514221.3526148
• 发表时间: 2022
• 期刊: SIGMOD
• 作者: Milkai, Elena;Chronis, Yannis;Gaffney, Kevin P.;Guo, Zhihan;Patel, Jignesh M.;Yu, Xiangyao
• 通讯作者: Yu, Xiangyao